Synopsys, Inc. (SNPS) Earnings Call Transcript & Summary

Plural, yes. That could be another 30 minutes discussion. Unfortunately, time -- I still have a few more things, a few other things I want to hit. So maybe let's move on to the next topic. I want to look a little bit beyond 2023. Well, I recently very equally thrilled to have heard a presentation from one of your team members, well, at an industry even actually, that was one hosted by Samsung. He talked about 3 things: that EDA can add value to the semiconductor industry are one, of course, managing the design complexity you just mentioned, right? And also the energy consumption of chips, I think last night, TSM also mentioned why we're moving to N2M 1.4. It's not just about shrinking the size, a lot more important is energy consumption. But meeting the need -- the second is meeting the need, the second is meeting the need workload optimize the chips. I really like that [indiscernible] workload optimize the chips. That's what the custom ASICs is all about. Three, but I think I like this point, three is about adding value to the talent and productivity issues for the industry that is special. I really want to pick this one and get your thoughts on talent and productivity. So we've heard you talk about implementing artificial intelligence to do chip design, which may be a great way for the industry to improve the productivity of chip design teams when -- well, quite honestly, we still have a shortage of design engineers as last time I check, but also accelerating the time to market. So my question really is this. I understand the history of EDA is all about improving productivity. That's where the name comes from, right, the design automation. But what -- what do you is new this time? And what it really excites you this time when it comes to like implementing AI into chip designs. Is this an evolutionary step in the history of EDA? Or this is something really revolutionary. Maybe some of us may have failed to see it yet.

Yu Shi, I love the way you've closed the question because I've always been asked is EDA revolutionary or evolutionary, and the way I would answer is we do revolutionary things and package them in evolutionary adoption. And the reason you need the evolutionary is, if you forget a single lesson learned in the past, no chip will work. And so it's accumulation of everything we've learned and don't screw that up and then some big leaps forward of automation or new ways of doing things. And yes, I am super excited about this because as of 2 years ago, suddenly, we saw the breakthrough of AI having a major impact on the overall productivity the way you defined it, of design. And as a personal anecdote, I want to give you the sense that when starting Synopsys, literally 36 years ago, we had that exact same feel with Synthesis. You could take some design. And in a matter of few hours, make it 30% better, 30% smaller, faster and smaller. And initially, people couldn't believe it and the graphs looked exactly the same. The algorithm would work, work, work and the results will get better and better. And so here we are now, 36 years later, or actually 34 years later because we started about 2 years ago. And for the first time, I saw graphs of progress that looked exactly the same in terms of better meet a bit better, better, better in a fraction of the time -- in a fraction of the effort time, right? And I instantaneously realized, "hey, we finally have sufficient computation, the right algorithms to make it happen and what we will be able -- what the first goal will be is, can the design be as good as what a human can do, but can we do it in -- and in this case, going to turn out to be 1/3 of the time. So 3 months to into 3 weeks. That's sort of the effort. And so then the second thing is, well, if we can do that, can we also make the results a little bit better? And the answer is, yes, we can also do that. And better means mostly 2 things, which is faster speed, lower power and you mentioned power. Power is going to be [ V ] variable for mankind in every which way you can think about it as a challenge. So that push is going to be very hard. And what is so fascinating is that we're applying this not to individual tools. We are applying this to entire piece of design flow. And that's why it does have impact on how much can leverage can you get off your engineers because just like 35 years ago, there were like 3% of the engineers said, hey, you're taking my job away, automating. But the other 97 said, now, I can do so much more. And we have the same reaction now. And I can substantiate that by the fact that we have now a large number of people have already adopted it and some that have already put it into massive production use because the benefits are so big. And still, I think we're just at the beginning of a couple of decades of advances. So that's pretty exciting.

Okay. Revolutionary, well, progress, that looks like a revolutionary over decades, actually may look revolutionary in a short period of time. But let's not forget about [indiscernible] small steps. Okay. So I think there are a few emerging opportunities for EDA that you have talked about over the past few years, quarters, well, hyperscalers, automotive, AI start hubs, among others, which are very interesting new verticals for EDA and I want to ask you about something else. I want to ask you about silicon photonics. I was formally a fab guy, I worked at Applied materials. I know you're a board member, sitting on the board. But I mean, I have a little bit of bias towards the more traditional ICs, more silicon ICs, but I can use a little bit of help here from you on silicon photonics. For one I think last year, you formed a JV called OpenLight with the Juniper Networks, photonics IPs and you also got your foundry partners, right? I realize. And I know your team said it's probably not material to our financials in the near term. But well, you just said, well, the revolutionary progress probably look a little bit evolutionary at the beginning, right? So I want to know what do you think the trend on silicon photonics is here. There are a lot of speculations on this, I mean, over the past year, including some articles claiming that [ NVIDIA ] have delivered, it's one of your top customers, they may be working with TSMC on some silicon photonics-based multi-GPU kind of solution that's another kind of systems of chips, right? So compared with -- compared to like traditional ICs, what is Synopsys role here and what the role does OpenLight play in your overall strategy on silicon photonics?

Okay. Well, the first perspective is, of course, why is silicon photonics of any interest? Because Photonics has had fundamentally the capability of transmitting data at very, very high speeds, right? And this is not new. But the words silicon photonics makes it new because the attempt is to say, "do the computation or the regular silicon task and on the same silicon, CEF1 can get a photonics device so that the communication can go out at high speed." And you say, "Well, why don't you just do that?" Well, you don't just do that is because each one of those require different manufacturing steps and bring those together have been economically and technically -- technically difficult and economically not satisfactory. And so as we now go to a whole next generation where I mentioned earlier, big data is truly big data and the speed of connectivity is absolutely the issue between chips or between racks or between in the space of a whole compute environment. there are a lot of efforts to bring these things together. And you can say, well, you'll put them on one of those multi-die things. But then do you want the dies to be connected at a super high speed. These are all somewhat open questions. And when I say open, I always mean techonomic. Is it technically feasible? And is it economically viable? And while we don't really give any specifics out of what OpenLight is doing, it's all part of an effort to drive these technologies forward. So that when the moment comes, we have -- and we have that, I think, already today, the EDA tools, some of the IP pieces ready to go and silicon photonics has proven to be one of those. Next year, it's going to take off. Next year it's going to take off. And the thing is the regular silicon world has really kept up very well with speed. But in the long term, there's no question that it will matter. And so maybe if I can broaden that statement to the next level, which is, I think, with the multi-die, you're going to see not just digital, memory, compute and stacking happening, but you're going to see other pieces come into the picture. Analog mixed signal, maybe some sensors, maybe some power devices. It's all about how much can you cram together. And so connectivity is one of the pieces of that equation. And so anything that can, at some point in time, revolutionize things, meanwhile, needs to be evolutionary prepared so it fits.

Great. Great. So I think I really want to make sure we have time to talk about software integrity. Well, if I'm completely honest. Well, again, I was a fab guy. I really had a tough time to understand EDA before I started covering Synopsys. But on top of that, I mean, SIG, Software Integrity Group, it's -- so it's slightly more difficult subject for someone like me with a semi background. So maybe I should have more. But I did hear a wide range of views from investors on SIG, from someone who really my question, whether it's a distraction to your core EDA IP business? Or to someone who love that you guys have the optionality and a very distinctive growth vector relative to your peers, I mean -- and someone who really admire you, I mean, I did here. And I think you're creating a new industry here, and just like you did over the last past 30, 40 years to EDA. Well, mind if you kind of walk us through the strategic rationale here over the years since we know that this business was kind of established around 2014 when you acquired a company called Coverity over the near term? And I have a near-term question on SIG as well. Yes. Why don't we start with the strategic rationale question, Yes.

Okay. Let me give you the historical rationale and then the evolved rationale. The historical rationale was that already by mid-2000, 2006, '07, I diagnosed that in semiconductor companies, they have more than half of their engineers were software engineers. And you say, well, that may be surprising. But the fact what it told me a couple of things. A, with some slowdown of Moore's Law, more of the functionality was in the software plus software had the benefit of being more flexible, make changes than hardware has, right? Secondly, it told me that, hey, better watch out because there's an extremely high discipline on designing hardware because if you make a mistake, and that you would know well, you're going to have to pay a mask set and remanufacture that's expensive, takes a lot of time. Whereas software, we just can send out a patch. Now that works, but it's not a great approach because if the software is of poorer quality than the hardware, then you keep iterating. And so when we acquired Coverity, it was in the full knowledge that -- by the way, half of their business was semiconductor related for embedded software and that this was a pathway towards higher quality discipline in software, and that was their first focus. At the very moment that we concluded that something else happened, which is security data to the picture as a rapidly growing catastrophe for so many situations. And so -- and the biggest vulnerability was surprise, surprise, actually, the software. And so quality and security were sort of de facto becoming the same. And that had another benefit in our vision, which is that we have already recognized with smart everything what the semiconductor industry would provide would actually go into all of these different vertical segments. And some of those -- pretty much all of them have need for security. But some of them, it's security moving to safety, right? And so if you can hack a car, you say, well, I know bad luck. Now you can say that may be extremely dangerous. And so fields where safety was important. They, in turn, started to focus on to security of the software. And it is only a matter of time where the security of the software and the security of the hardware are interwoven. And actually already today, Synopsys has actually a quite active slice of business in our IP collection to actually embed security modules, routes of trust and things like that in there. And so now you can start getting a sense of the long-term perspective, it was a move up that value stack where at the end of the day, everything is connected to everything else and being able to measure things all the way in the hardware or all the way in the software on the other one is absolutely imperative. Now the SIG business, Software Integrity business meanwhile has grown beyond semiconductor substantially because all these other companies also need security, and there's a ton of companies develop a lot of software. And so it's been actually quite intriguing to suddenly interact with companies that we would never had anything to do with shopping stores, Shell oil financial entities, health companies. And in that context, while it is a TAM that is sort of removed traditionally from semiconductors. Over the long term, I see that there will be gradually also more connections between the two. Hopefully, that gives you a little bit holistic perspective.

Yes. That's great. I think we probably have a little bit of time, but unfortunately, I have to throw in some near-term question that's always [indiscernible] for our investors and probably well to satisfy their need. I will ask you a near-term question on SIG. I know you sort of said on your last earnings call, you expect all segments of your business to grow similarly at mid-teens into your fiscal '23. But SIG business there's a concern that maybe SIG is not as resilient in terms of keeping up the growth rate as the EDA IP part of the business. Is that right or wrong and any thoughts, that would be great.

Yes, the comment came from both an observation and an understanding of markets, which is that -- the market that we touch with SIG is very, very broad with many, many companies all over the spectrum of what companies or anybody who has software, sooner or later has some of these issues. And when you have economic uncertainty, and certainly, I would say the fall was maybe a little bit more pronounced now it goes up and down. But they were -- at that point in time was inflation, inflation, inflation, right? And then the question, does that bring a recession or does that just mean a readjustment of the market because certain people will play pricing, which makes it worse. But what is clear is that when there's uncertainty, top managements of company says, okay, what do we absolutely have to have and what is nice to have. And so any investment that tend to go towards long-term quality and so on would be viewed as prophylactic, right? It's sort of like it's really hard to sell vitamins where people are not yet sick and they have no money. Whereas if you have a headache, it's really easy to sell an aspirin than anybody who's ever had that knows that in an airport, single aspirin $4.50, right. And so why am I bringing it up? Because when you're in the business of prophylactically help people design better software, of course, economically over time, has a big impact. But in the moment, people can just say, well, let's slow that down. And initially, during all these discussions of the market may be softening. We saw a little bit of that. And the way you see it's mostly that the closure of transactions takes a bit longer. They still closed. And so I don't think that the market has changed. I think the opportunity is just as rich as before. It's more what are the things that are a little bit more time sensitive than others in the decision-making. And by the way, we have made that business already much more solid because we're in multiyear agreements there. And the rest of Synopsys or the EDA part of Synopsys has now for quite a number of years, have very strong ratable business and gradually, I think, is becoming like that, too. So that was just full disclosure in the context of the question marks at that time, and we'll see how it works. But so far, there's no indication that there's an issue.

Thank you, Aart. So we got 3 minutes left. I think let's take one question from the audience. There's an interesting one. Well, I think I can tell that people really look up to you, Aart, and try to ask you about things at a very high level. So the question is, what are the hurdles to really get to a system of chips world. I mean the true systems chips world is the manufacturing side, advanced packaging the hurdle? Or what's your view here?

Well, so my view is that we have entered a very different phase of the industry because from the tradition, what I would call scale complexity, which is more transistors, or more transistors, or more transistors, we're now in systemic complexity, which is different problems intersecting. And so the minute you have multiple chips, you have a whole different world, which is how do you connect them. Can you stack them on top of another. Well, stacking sounds good, right, that's as close as they get, except if the one underneath is a heater, the guy above may not like that, right? But okay, so now you have to deal with that issue. How do you deal with the fact that when you have multiple chips on a super chip together and something doesn't work, which chip is it? Whose fault is it? Is it the same technology? Is it a different technology. And so we've entered an age of systemic complexity. And I would argue, we've seen visible notion of that or just supply chain is systemic complexity. But now building these things is also the next level of systemic complexity. And therefore, our main mission at Synopsys is, in simple terms, make it all work. That means we need to understand testing, verification, connectivity, heat transfers, make it all work. That's a fabulous assignment, and I think we're well positioned because we have many of the pieces to do exactly that. And so we'll make it happen.

Yes. Great. I think that's a great way to close today's fireside chat down. Well, I think that the answer you just provided. Well, I kind of feel like while complexity is EDA's friend, I mean, you love complexity. And well, time really flies. And I mean, we'll probably want to end the session right now, but I really want to thank Aart, and thank Lisa, who is in the background for your time with us today and really appreciate all the insights especially coming from Aart and I look forward to catching up with you again after your fiscal Q1 earnings call. That's a wrap, everybody, and enjoy the rest of your conference.

Thank you, Charles, for the very interesting questions.

Thank you.